Abstract: The core factor affecting the performance of A₀ modal Lamb wave touch screen is the quality of the acoustic fingerprint database. The quality evaluation and optimization of the acoustic fingerprint database are studied through statistical and analytical method of Manhattan distance. Firstly, 900 layout schemes of excitation-receiving sensors are simulated with finite element model and 900 corresponding databases are collected. Then, a set of nephograms of Manhattan distance and its frequency distribution for each database are constructed. The quality of each database is evaluated according to the features of the nephogram set and the weighted sum values of the frequency histogram. The layout scheme of excitation-receiving sensors is optimized according to the database quality. Finally, the influence of excitation signal on the quality of acoustic fingerprint database is analyzed. Compared with the tone burst excitation signal, the broadband chirp signal can significantly improve the database quality. The above researches provide a basic method to optimize the quality of acoustic fingerprint database for Lamb wave touch screen.